At its I/O 2025 developer conference, Google announced a new feature in Google Meet that enables real-time speech translation using its Gemini AI model.
This update is designed to bridge language barriers during virtual meetings by translating spoken language almost instantly, while preserving the speaker’s original voice, tone, and emotion.
The feature works by using Gemini’s advanced generative capabilities to create near-real-time dubbed translations that mirror the speaker’s vocal qualities.
Unlike simple subtitles or robotic text-to-speech, this translation method captures vocal inflexions and emotional nuance, creating a more natural and human-like listening experience.
In a live demonstration, an English speaker and a Spanish speaker held a conversation using Google Meet. As the Spanish participant began speaking, the tool immediately translated and dubbed their speech into English, maintaining their voice’s rhythm and tone.
This two-way translation allows both parties to interact fluidly in their native languages, significantly enhancing communication without relying on external interpreters or subtitles.
Initially, the feature supports English and Spanish translations. Google plans to expand this to include Italian, German, and Portuguese in the coming weeks.
The feature is currently being rolled out in beta and is available to users subscribed to Google’s premium AI plans. Specifically, it is offered to users of the AI Pro and AI Ultra plans, with the latter priced at $249.99 per month.
This move brings Google Meet in line with competing services like Microsoft Teams, which also launched a similar real-time translation tool earlier in the year.
However, Google’s emphasis on preserving the emotional fidelity of speech sets its tool apart from more mechanical-sounding alternatives.
Google has not yet disclosed when this feature will become generally available, nor whether it will expand to more languages beyond the planned additions.
